基于谱峭度的风电机组传动链故障特征提取方法研究

发布时间：2018-06-28 14:19

本文选题：风电机组 + 谱峭度　；参考：《华北电力大学(北京)》2017年硕士论文

【摘要】：风电机组具有载荷复杂、运行环境恶劣等特点,因此故障率和运维费用高。其中,风电机组传动链的问题尤其突出。针对风电机组传动系统的特点,研究新的故障特征提取理论和方法,提高故障早期诊断能力,对于保证风电机组的健康运行具有重要的意义。在对风电机组实际振动信号进行诊断分析时,经典的振动信号处理方法难以捕捉到反映故障特征的准确信息,影响了对机组故障的诊断分析。本文针对以上问题,研究基于谱峭度的振动信号分析和故障特征提取方法,结合最小熵解卷积(Minimum entropy deconvolution,MED)和最大相关峭度解卷积(Maximum correlated kurtosis deconvolution,MCKD)对振动信号进行分析,从强背景噪声下的振动信号中准确的提取故障特征信息,并应用风电机组传动链的实测信号进行验证。主要研究内容和结论如下:(1)对谱峭度理论基础进行探究,介绍了两种传统谱峭度算法,并用仿真信号及风电机组实测信号对比分析,探讨了谱峭度算法的特点及不足。(2)针对强噪声环境下,谱峭度算法存在故障特征难以提取等问题,研究了将谱峭度与AR-MED、MCKD结合提取微弱故障特征的方法。首先对信号进行滤波预处理,抑制噪声成分,提高了冲击特性,然后再对降噪后的信号进行谱峭度计算,包络解调提取出故障特征频率。用风电机组实测振动信号对该方法进行检验,结果表明早期故障特征能力有所改善。(3)研究用振动特征值表示风电机组振动状态劣化趋势的方法,对数据进行等时间间隔筛选,运用改进的谱峭度方法提取故障特征,做出故障特征值变化趋势,通过与常用的时域特征值变化趋势进行对比,表明改进谱峭度方法的特征值能够反映故障的发生发展趋势。
[Abstract]:Wind turbine has the characteristics of complex load and bad operating environment, so the failure rate and operation cost are high. Among them, the problem of wind turbine transmission chain is particularly prominent. According to the characteristics of wind turbine transmission system, it is of great significance to study the new theory and method of fault feature extraction to improve the ability of early fault diagnosis for ensuring the healthy operation of wind turbine. When the actual vibration signal of wind turbine is diagnosed and analyzed, it is difficult for the classical vibration signal processing method to capture the accurate information reflecting the fault characteristics, which affects the diagnosis and analysis of the unit fault. In this paper, the vibration signal analysis and fault feature extraction based on spectral kurtosis are studied in this paper. The vibration signal is analyzed by minimum entropy deconvolution (MED) and maximum correlated kurtosis deconvolution (MCKD). The fault characteristic information is extracted from the vibration signal of strong background noise and verified by the measured signal of the transmission chain of wind turbine. The main research contents and conclusions are as follows: (1) the theoretical basis of spectral kurtosis is explored, two traditional spectral kurtosis algorithms are introduced, and the simulation signals and wind turbine measured signals are compared and analyzed. The characteristics and shortcomings of spectral kurtosis algorithm are discussed. (2) aiming at the problem that the spectral kurtosis algorithm is difficult to extract fault features in strong noise environment, the method of extracting weak fault features by combining spectral kurtosis with AR-MEDMKD is studied. Firstly the signal is pre-processed to suppress the noise component and the impulse characteristic is improved. Then the spectrum kurtosis of the de-noised signal is calculated and the fault characteristic frequency is extracted by envelope demodulation. The method is tested with the measured vibration signal of wind turbine, and the results show that the ability of early fault characteristic is improved. (3) the method of showing the deterioration trend of wind turbine vibration state with vibration eigenvalue is studied, and the data are screened at equal intervals. The improved spectral kurtosis method is used to extract the fault feature and make the trend of fault eigenvalue change. The comparison with the time domain eigenvalue shows that the improved spectral kurtosis method can reflect the trend of fault occurrence and development.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM315

【参考文献】